From: Andrea Manca [SMTP:[EMAIL PROTECTED]] IMPUTE: request for codes for MI in STATA Hello everybody (delurk). As I have understood it, the problem with using Schafer's methods is implementing the Markov Chain Monte Carlo (MCMC) algorithm. To my knowledge, this would be difficult using Stata. Most people implementing new Stata functions program some basic stuff and throw it at Stata's optimizer. Stata's not made for more advanced tasks. With MI with MCMC you actually need to assess convergence yourself graphically. This is one of the barriers to its acceptance. (Right?) I'm in your situation. I use Stata, but bought S-PLUS to do MI. But it seems I'm the only one in Norway who has realized the benefits of MI. So I'm slightly intimidated by "going it alone", especially in assessing convergence and making sure that Schafers software does what I want it to do. Is this list appropriate for basic questions ? Yours, Jan Brogger PhD student, Respiratory Epidemiology Group, University of Bergen, Norway
